我在做内核编程。 我想知道在内核模块中使用vfs_rename（...）函数之前要保留哪个mutex_locks（）。
vfs_rename的原型（...）
int vfs_rename（struct inode * old_dir，struct dentry * old_dentry， struct inode * new_dir，struct dentry * new_dentry，）
由于
答案 0 :(得分：1)
vfs_rename有评论：
4089 /**
4090 * vfs_rename - rename a filesystem object
4091 * @old_dir: parent of source
4092 * @old_dentry: source
4093 * @new_dir: parent of destination
4094 * @new_dentry: destination
4095 * @delegated_inode: returns an inode needing a delegation break
4096 *
4097 * The caller must hold multiple mutexes--see lock_rename()).
4098 *
4099 * If vfs_rename discovers a delegation in need of breaking at either
4100 * the source or destination, it will return -EWOULDBLOCK and return a
4101 * reference to the inode in delegated_inode. The caller should then
4102 * break the delegation and retry. Because breaking a delegation may
4103 * take a long time, the caller should drop all locks before doing
4104 * so.
4105 *
4106 * Alternatively, a caller may pass NULL for delegated_inode. This may
4107 * be appropriate for callers that expect the underlying filesystem not
4108 * to be NFS exported.
4109 */
因此，它将我们带到lock_rename函数：
2440 /*
2441 * p1 and p2 should be directories on the same fs.
2442 */
2443 struct dentry *lock_rename(struct dentry *p1, struct dentry *p2)
2444 {
2445 struct dentry *p;
2446
2447 if (p1 == p2) {
2448 mutex_lock_nested(&p1->d_inode->i_mutex, I_MUTEX_PARENT);
2449 return NULL;
2450 }
2451
2452 mutex_lock(&p1->d_inode->i_sb->s_vfs_rename_mutex);
2453
2454 p = d_ancestor(p2, p1);
2455 if (p) {
2456 mutex_lock_nested(&p2->d_inode->i_mutex, I_MUTEX_PARENT);
2457 mutex_lock_nested(&p1->d_inode->i_mutex, I_MUTEX_CHILD);
2458 return p;
2459 }
2460
2461 p = d_ancestor(p1, p2);
2462 if (p) {
2463 mutex_lock_nested(&p1->d_inode->i_mutex, I_MUTEX_PARENT);
2464 mutex_lock_nested(&p2->d_inode->i_mutex, I_MUTEX_CHILD);
2465 return p;
2466 }
2467
2468 mutex_lock_nested(&p1->d_inode->i_mutex, I_MUTEX_PARENT);
2469 mutex_lock_nested(&p2->d_inode->i_mutex, I_MUTEX_CHILD);
2470 return NULL;
2471 }